Reading Guide to: Hindess, B (1981) 'The Politics of Social Mobility: A Review Symposium', in Economy and Society, Vol 10

This is a critical review of the Oxford Mobility Survey (see file), and it also illustrates the critical power of Hindess's approach to discourse analysis.

Goldthorpe's main concern was with the class formation and its tendency towards openness or closure, and this produced his criticisms of liberal ideology. Basically, an open society only becomes possible through class struggle, and this helps focus the political implications of social mobility for class formation [one main conclusion was that the working class had remained relatively solid, while the service class had largely absorbed its new entrants -- only the intermediate classes were fragmented]. Halsey's criticism of liberalism, together with his view that further expansion of the education system will produce equality, also led to a break with liberal ideology. However, one issue that arises immediately is whether empirical investigations and political concerns can be connected quite so smoothly.

Halsey in particular attacks empiricist approaches to social mobility, but he is still in the trap of empiricist thinking. Why use age - cohort analysis, for example, and why only consider men? There seem to be two answers: (a) a technical one, pointing to the need to replicate other famous studies, especially Blau and Duncan; (b) an empirical one, referring to the family structure as prioritising men and their social mobility. To analyse social mobility properly would require much larger cohorts, and the team lacked sufficient resources; accepting women as well as men would also lead to undersized cohorts [for comparative purposes?]. However, why was it so important to tie this study so closely to Blau and Duncan in the first place?

Blau and Duncan produced the first attempt to use path analysis to measure the effects of ascribed, achieved, and intervening variables. However, they confused statistical relations with causal factors, and the variables they chose to study seemed to be rather odd constructs. 'Family climate' is a good example, but their notion of class is equally odd. In general, statistical relations are never just simple labels for social relations, but can themselves be the product of 'structural features of social relations'. For example, the statistical relation between sex and achievement arises because the education system discriminates against women, not that sex somehow causes (under-)achievement. Even if we accept the strategy of the study, a number of different causal models are possible -- there are large residuals in Blau and Duncan [that is, amounts of variance unexplained by the chosen variables].

Halsey himself reproduces some aspects of the Blau and Duncan study, in measuring ascribed and achieved status for example, and in trying to isolate the effects of families on school. There are some similar problems as a result:

(a) There are very high residuals in the study, especially in trying to demonstrate the effects of class rather than culture. This has led Ridge to speculate that variations of types of secondary school are also important. [Compare this with similar criticisms of the Plowden study too -- see file]

(b) Statistical relations are taken as causals again. Thus there is a view that social class causes early leaving, instead of an analysis of the actual mechanisms of discrimination inside schools.

(c) Halsey is 'cavalier' in his references to Bourdieu -- 'family climate' is much wider than 'cultural capital' alone, and Halsey's attempt to operationalise Bourdieu is a classic example of methodological empiricism.

Goldthorpe has more doubts about Blau and Duncan, especially their focus on individual attributes and status. This is a unexamined split in the Oxford Team. Goldthorpe never specifies any precise doubts about path analysis, but he does take a different approach to Blau and Duncan, focusing on the consequences of mobility rather than explanations of it, and the concomitants of social mobility rather than any causals involved. Hindess thinks this strange, because anyone interested in politics should be concerned with the causals involved. Given these criticisms, why should the Blau and Duncan design still be seen as suitable at all? [It seems to be the old story of scholastic interests dominating political ones in the end?]

Turning to politics, both Halsey and Goldthorpe are critics of liberal theory. Goldthorpe in particular uses the concept of 'absolute' versus 'relative' mobility to criticise liberal expectations of the effects of economic growth. [Basically, economic growth produces large absolute amounts of social mobility, but improves the chances of working class people only a little relative to their competitors in the other classes]. This leads him to speculate about what real egalitarianism might look like, and to express doubts about piecemeal legislation. Liberals have always underestimated the resistance of the social class system to reform, driven by attempts to preserve privilege. Goldthorpe also has a general concern for the social consequences of class formation [not only the effects on solidarity or 'density', but some of the personal consequences of becoming socially mobile]. However, this political commitment reveals a major discrepancy with the actual study. In particular, there are no empirical investigations of class struggles, nothing on the role of ideologies, and nothing to guide calculative British politics [Hindess's preferred form of politics, having dispensed with marxist theory as some scientific guide to 'real' politics after his marvellous theoretical efforts -- see his other file].

We're left with general neo-Weberian politics [and Parkin is mentioned here]. This concerns itself with conditions of solidarity and closure [one of the reasons for investigating the effects of social mobility on other social connections, with kin or with friends]. Goldthorpe operates with classes as idealised political actors rather than investigating the actual specifics of trade unions or political parties. This produces an 'allegorical' class structure, where politics is reduced to relational and experiential aspects of mobility. This is a 'sociologisation of politics', where politics merely reflects some underlying 'real' social conditions. [Spot some of the parallels here with Coward's critique of Hall's views of the class system].

The exclusion of women from the study remains as a real puzzle -- why should the team want to replicate the Blau and Duncan study enough to feel obliged to leave women out? There is the 'family' argument [that the occupation of the male earner dominates the class location of the entire family with few exceptions -- much disputed since, of course], but Goldthorpe claims to be interested in the social mobility of individuals. Anyway, the class structure is defined as consisting of occupations which are not necessarily male ones. Finally, even male class structures operate through and are influenced by families, of course [men also give up chances of social mobility in favour of 'family matters' and preferences, and can also gain chances from marriage etc?]. Two conceptions seemed to be involved in the choice to omit women -- that occupational positions exist regardless of sex, and that the structure of families is some independent [and unchanging] variable.

The argument that males are heads of families is a generalisation, and there may well be significant differences, in income say, in families where women do paid work as well. Many of the improvements in the standard of living [with important social and political consequences] would not be possible without the development of women's work, but this is not investigated at all. This seems to be some argument that there is a residual functional division of sex roles in families [and Hindess finds further evidence for this in additional work by Halsey on families]. There are cross-class and other kinds of mobility open to women, including both direct and indirect paths such as mobility via marriage. None of these are investigated in the study, yet they could all be crucial for political action.

The above represents perhaps the most massive discrepancy between the political concerns of the authors and the empirical study itself. As a result of this discrepancy, the Oxford Survey remains as a 'technically competent irrelevance'. It helps to dispose of false preconceptions [liberal ones especially], but has nothing to say about the conditions of political action.